low american life expectancy 1970-2009, insights from a new contour decomposition method alyson a....
TRANSCRIPT
Low American life expectancy 1970-2009, Insights from a new contour decomposition method Alyson A. van Raalte, Vladimir M. Shkolnikov, and Dmitri Jdanov
Introduction
Results
Approach
18.04.23
Summary and Interpretation of Results
What is known about the low life expectancy at age 50 of USA compared to other highly developed countries?•Cross-sectional studies: USA has comparatively high early adult mortality and low old age mortality•Trend studies: The USA experienced comparatively strong heart disease mortality reduction, but weak declines from other causes of death, notably strokes, cancers and mental disorders over the past half-century
Initial levels of mortality bear little correlation with mortality trends
Current study: We quantify the effects of past differences in the age-specific mortality levels (1960) and differences in age-specific mortality trends (1960-2009) to recent life expectancy differences (2009) between 18 highly developed countries. This is the first time trends and levels are compared on the same scale.
X
Females: all-cause mortality
X
Y Males: all-cause mortality
Japan
France
Spain
Australia
Italy
Switzerland
Canada
Finland
Sweden
Norway
Austria
Netherlands
Belgium
New Zealand*
UK
Ireland
Denmark
50
-59
60
-69
70
-79
80
-84
85
+
Age
Initial difference 1960
50
-59
60
-69
70
-79
80
-84
85
+
Trend 1960-2009
50
-59
60
-69
70
-79
80
-84
85
+
Total difference 2009
Australia
Switzerland
Japan
Canada
New Zealand*
Sweden
Italy
Norway
Spain
France
UK
Netherlands
Ireland
Austria
Belgium
Finland
Denmark
50
-59
60
-69
70
-79
80
-84
85
+
Initial difference 1960
50
-59
60
-69
70
-79
80
-84
85
+
Trend 1960-2009
50
-59
60
-69
70
-79
80
-84
85
+
Total difference 2009
-2.86
-1.75
-1.25
-0.75
-0.25
0.25
0.75
1.25
1.75
2.25
Contribution to USA life expectancy
difference (years)
Figure 1: Decomposition of the 2009 difference in remaining life expectancy at age 50 between the United States and other high-income countries, arranged in order of increasing e50 in 2009. Each square represents the age contribution to this difference for each country: the total difference is the sum of the initial difference and the trend components. Orange-red hues are to the advantage of the USA while blue colours are to the advantage of other countries.
X
Females: absence of smoking
X
Y Males: absence of smoking
Japan
France
Australia
Spain
Canada
Italy
Switzerland
Finland
Sweden
Norway
Austria
New Zealand*
Netherlands
UK
Ireland
Belgium
Denmark
50
-59
60
-69
70
-79
80
-84
85
+
Age
Initial difference 1960
50
-59
60
-69
70
-79
80
-84
85
+
Trend 1960-2009
50
-59
60
-69
70
-79
80
-84
85
+
Total difference 2009
Japan
Canada
Australia
Italy
Netherlands
Spain
Switzerland
France
New Zealand*
UK
Belgium
Norway
Ireland
Sweden
Austria
Denmark
Finland
50
-59
60
-69
70
-79
80
-84
85
+
Initial difference 1960
50
-59
60
-69
70
-79
80
-84
85
+
Trend 1960-2009
50
-59
60
-69
70
-79
80
-84
85
+
Total difference 2009
-2.72
-1.75
-1.25
-0.75
-0.25
0.25
0.75
1.25
1.75
2.25
Contribution to USA life expectancy
difference (years)
Figure 2: Decomposition of the 2009 difference in remaining life expectancy at age 50 in the absence of smoking between the United States and other high-income countries, arranged in order of increasing e50 in 2009. Smoking non-attributable mortality was estimated using the Preston, Glei & Wilmoth method (2010, 2011).
Contour-replacement decomposition (Jdanov & Shkolnikov 2014, MPIDR WP 2014-010)
a)Step-wise replacements from A2 to A1 to B1 to B2 for each ageb)After each replacement, calculate the difference in life expectancy.c)Reverse steps a & b, from B2 to B1 to A1 to A2d)Average elementary contributions
Result: two age-specific components: 1.differences in the initial age structure of mortality (orange arrow) 2.different age-specific mortality trends (the grey arrows)
The sum of these two age-specific components exactly equals the classic decomposition of e0 differences between A2 and B2 into age-specific components (yellow arrow).
Data come from the Human Mortality Database and the WHO mortality database
Illustration of the step-wise decomposition algorithm
Conclusion
• Most of the current gaps in life expectancy between USA and others lie in past age-specific mortality differences (Fig 1)
• Men: American men experienced above average mortality reductions over young working ages (50-70), but worse trends at older ages.
• Women: American women had comparatively weak mortality reductions at all ages.
Possible Explanations• Smoking: Controlling for smoking attenuated but did not alter this pattern
(Fig 2)• CVD: Stronger reduction in USA – in absence of CVD Americans had
stronger initial advantage and weaker trends (Fig 3)• Obesity: USA declines comparatively stronger at younger than older ages.
Obesity unlikely to be major contributor. • Health care system: Weak-trend countries (Scandinavian countries,
Netherlands, USA) have very different health care systems. Notable that American men had strong mortality declines over ages not covered by Medicare, but weak declines at old ages, where American health care is thought to outperform other countries (NRC report 2011).
• Selection: Consistent with high young, low old age mortality – but attenuation of pattern would mean that USA is getting relatively less heterogeneous compared to other countries. Any evidence of this?
• Data issues: Age misreporting that is weakening over time consistent with weakening USA advantage at old ages. Unlikely to explain the high early adult mortality.
X
Females: absence of CVD
X
Y Males: absence of CVD
Japan
Spain
France
Australia
Italy
Switzerland
Finland
Austria
Canada
Sweden
New Zealand*
Norway
Belgium
Ireland
Netherlands
UK
Denmark
50
-59
60
-69
70
-79
80
-84
85
+
Age
Initial difference 1960
50
-59
60
-69
70
-79
80
-84
85
+
Trend 1960-2009
50
-59
60
-69
70
-79
80
-84
85
+
Total difference 2009
Australia
Sweden
Switzerland
New Zealand*
Canada
Italy
Japan
UK
Norway
Finland
Spain
Ireland
Austria
Netherlands
France
Belgium
Denmark
50
-59
60
-69
70
-79
80
-84
85
+
Initial difference 1960
50
-59
60
-69
70
-79
80
-84
85
+
Trend 1960-2009
50
-59
60
-69
70
-79
80
-84
85
+
Total difference 2009
-4.36
-1.75
-1.25
-0.75
-0.25
0.25
0.75
1.25
1.75
3.26
Contribution to USA life expectancy
difference (years)
Figure 3: Decomposition of the 2009 difference in remaining life expectancy at age 50 in the absence of CVD between the United States and other high-income countries, arranged in order of increasing e50 in 2009. Mortality in the absence of CVD was estimated using traditional cause-deleted life table methods.
• No single theory or mortality determinant was satisfactory in explaining these international sex-specific divergences.
• Too often, the poor standing of Americans in life expectancy tables is linked to contemporary social problems, without any reference to longstanding differences.
• Our results make clear that any discussion of American shortfalls in life expectancy must first begin with an examination of the past
*Data in the later period are from 2008.
*Data in the later period are from 2008.
*Data in the later period are from 2008.